US8694316B2 - Methods, apparatus and computer programs for automatic speech recognition - Google Patents
Methods, apparatus and computer programs for automatic speech recognition Download PDFInfo
- Publication number
- US8694316B2 US8694316B2 US11/255,201 US25520105A US8694316B2 US 8694316 B2 US8694316 B2 US 8694316B2 US 25520105 A US25520105 A US 25520105A US 8694316 B2 US8694316 B2 US 8694316B2
- Authority
- US
- United States
- Prior art keywords
- user
- recognition
- phones
- input
- phone
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active, expires
Links
- 238000000034 method Methods 0.000 title claims description 33
- 238000004590 computer program Methods 0.000 title description 4
- 230000005236 sound signal Effects 0.000 claims abstract description 22
- 230000004044 response Effects 0.000 claims description 27
- 238000012549 training Methods 0.000 claims description 25
- 230000004048 modification Effects 0.000 claims description 3
- 238000012986 modification Methods 0.000 claims description 3
- 238000011156 evaluation Methods 0.000 abstract description 3
- 230000001186 cumulative effect Effects 0.000 description 16
- 238000009826 distribution Methods 0.000 description 11
- 230000003993 interaction Effects 0.000 description 11
- 230000007704 transition Effects 0.000 description 11
- 238000012790 confirmation Methods 0.000 description 8
- 230000008569 process Effects 0.000 description 8
- 238000012545 processing Methods 0.000 description 6
- 230000006978 adaptation Effects 0.000 description 5
- 230000000875 corresponding effect Effects 0.000 description 5
- 238000004364 calculation method Methods 0.000 description 4
- 230000001419 dependent effect Effects 0.000 description 3
- 238000013459 approach Methods 0.000 description 2
- 239000000284 extract Substances 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 235000015110 jellies Nutrition 0.000 description 2
- 239000008274 jelly Substances 0.000 description 2
- 235000004240 Triticum spelta Nutrition 0.000 description 1
- 230000003044 adaptive effect Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000001276 controlling effect Effects 0.000 description 1
- 230000002596 correlated effect Effects 0.000 description 1
- 238000002592 echocardiography Methods 0.000 description 1
- BTCSSZJGUNDROE-UHFFFAOYSA-N gamma-aminobutyric acid Chemical compound NCCCC(O)=O BTCSSZJGUNDROE-UHFFFAOYSA-N 0.000 description 1
- 229940084388 gammar Drugs 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 230000035755 proliferation Effects 0.000 description 1
- 230000000717 retained effect Effects 0.000 description 1
- 238000012552 review Methods 0.000 description 1
- 230000003595 spectral effect Effects 0.000 description 1
- 238000003786 synthesis reaction Methods 0.000 description 1
- 230000001960 triggered effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/08—Speech classification or search
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/02—Feature extraction for speech recognition; Selection of recognition unit
- G10L2015/025—Phonemes, fenemes or fenones being the recognition units
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/08—Speech classification or search
- G10L2015/085—Methods for reducing search complexity, pruning
Definitions
- the present invention provides methods, apparatus and computer programs for automatic speech recognition.
- An automatic speech recognition (ASR) system takes an audio signal as input and typically compares the input signal with known sounds (phones) and sequences of sounds (paths) of an acoustic model (AM) to identify words that appear to match a spoken sequence of sounds. After identifying a word or words corresponding to an input audio signal, a text or other machine-readable representation of the identified matching words may be returned by the ASR to an application program such as an interactive voice response (IVR) telephony application.
- IVR interactive voice response
- a confidence score may be returned with each apparently-matching word, the confidence score being based on the closeness of an incoming acoustic segment to the mean of a probability distribution associated with a phone within the ASR system's acoustic model. A number of possible words and their respective confidence scores may be returned for selection or further analysis.
- ASR automatic speech recognition
- microphone labels the sequence of sounds that the ASR system identifies must also be matched against an application-specific grammar, which predefines words and phrases that are expected. If the ASR system is trained on enough data and if the grammar covers all possible words and phrases, then recognition accuracy can be very high. However, individual sounds within a given language may be easily confused, such as “F” and “S” in English. Such sounds may well appear within the words in the application grammar. In such cases, recognition accuracy will tend to decrease.
- the ASR system may confuse the two items—in particular because a user prompt such as “Did you say Stephen James?” could be answered with “yeah”, “OK”, “correct”, or “nope”, “nah” and so forth.
- a first aspect of the present invention provides a method for controlling operation of an automatic speech recognition (ASR) system.
- ASR automatic speech recognition
- the ASR system During a ‘conversation’ with a user, the ASR system generates user prompts to elicit required inputs from the user. Sounds within an input audio signal are compared with phones within an acoustic model to identify candidate matching phones. Recognition confidence scores are calculated for the individual candidate matching phones, and the recognition confidence scores are evaluated to identify one or more of the candidate matching phones having a predefined recognition confidence characteristic (for example, identifying candidate phones having high or low confidence scores).
- the method includes selecting a user prompt for eliciting a subsequent user input, wherein the selection is dependent on the recognition confidence characteristic of the identified one or more phones.
- a typical ASR system in the context of this specification includes a speech recognition engine and an application program, and may include other components as described below.
- User prompts in the context of this specification are prompts that are presented to users to elicit user inputs. For example, a question may be presented to the user by playing an audio prompt (such as a stored audio clip or a dynamically generated, simulated voice message) or by displaying text on a computer screen. Each user prompt encourages the user to speak, for example to make selections from a menu of options or to specify the person or service that the user wants.
- sequences of phones that are identified as possible matches to the sounds of the input audio signal are compared with words in a speech recognition grammar (such as an application-specific grammar of an IVR application) to identify candidate matching words.
- a speech recognition grammar such as an application-specific grammar of an IVR application
- the confidence scores of the individual phones, within a sequence of phones that appear to match a word in the application grammar are combined to form a cumulative confidence score for evaluating the most likely word matches.
- a defined threshold low confidence
- the identification of candidate matching phones having predefined recognition confidence characteristics includes identifying phones having a recognition confidence above a defined threshold (high confidence) and prompting a user to input high confidence phones to avoid or overcome speech recognition problems.
- confidence scores can be used to determine which areas of language the ASR program was unable to recognize with high confidence. Confidence scores can also show which user prompts result in low recognition confidence scores because phone-specific confidence scores can be evaluated and recorded in association with the prompts that elicited the spoken inputs. Subsequently, a process for generating a user prompt takes account of the per-phone confidence scores as well as per-word (i.e. phone sequence) confidence scores to adapt its user prompts for improved speech recognition.
- the user prompts may be selected for the purpose of targetted speech recognition training—generating prompts which invite input of phones having low recognition confidence scores.
- user prompts may be selected to encourage users to input words or phrases other than the words that were difficult to recognize—attempting to avoid phones having low recognition confidence by inviting users to input alternative words.
- a first user input is required for a first application operation and a subsequent user input is required for a second application operation.
- One method according to the invention evaluates confidence scores associated with apparent matches between sounds within the first user input and phones within an acoustic model of the ASR system and then, in response to the confidence scores, controls the application's generation of user prompts associated with the subsequent required user input.
- the first and second application operations may be any two distinct operations of an application program, such as successive selection steps from an hierarchical menu in which the first and subsequent user inputs are used to indicate choices from the menu.
- one embodiment selects user prompts that are expected to encourage input of phones that were previously identified as having low recognition confidence scores—tailoring the ASR system's generated user prompts for targetted recognition training.
- the selection of user prompts can be implemented dynamically during the course of a normal user interaction with the ASR system, without requiring a dedicated caller-specific training period.
- User prompts can be modified for each successive step of a functional interaction between a caller and an ASR system. Because the caller is allowed to progress to the next successive step, this approach may be more efficient as well as less frustrating for callers than if the caller is asked to repeat an earlier input.
- One solution invites caller inputs that combine a word that includes a low confidence phone with one or more words that were previously recognised with high confidence.
- the ASR system selects a user prompt that is expected to encourage input of words having high confidence matching scores, or to discourage input of phones having low confidence matching scores.
- an application program within the ASR system may respond to a low-confidence recognition score from the ASR system's recognition engine by selecting a user prompt that specifies, as the required user input, a synonym or other alternative to a word previously recognized with low confidence. For example, if a caller's ‘Yes/No’ inputs are difficult to recognize (low confidence matching scores), the caller may be prompted to say ‘Correct’ instead of ‘Yes’ while keeping the remainder of the user prompt unchanged.
- low confidence scores may trigger replacement of a user prompt (such as “If ⁇ ConditionA>, say ‘Yes’” or “Which service do you require?”) with a more specific conditional instruction (such as “If ⁇ ConditionA>, say ‘Yes please’, and if ⁇ ConditionB> say ‘No’”) to invite a more specific response from callers.
- selection of a user prompt.
- a recognition grammar may be restructured or a recognition engine may be optimised in response to the evaluated recognition performance.
- Such changes to an application grammar may be based on phone confidence scores evaluated over a period of time and for multiple callers.
- the invention may be implemented to capture information about real and potential sound confusions and may use the information to modify and adapt the performance of the ASR system. Confidence score patterns can be used to generate recommendations for changing an acoustic model (by targetted training as described above), or changing an application grammar using a synonym dictionary.
- a second aspect of the invention provides an automatic speech recognition (ASR) system comprising an application program and a speech recognition engine, wherein the speech recognition engine includes program code for evaluating recognition confidence scores associated with apparent matches between input sounds and phones in an acoustic model.
- the ASR system controls the generation of user prompts for influencing user inputs to the application program.
- the application is preferably a speech-responsive application which selects user prompts based on recognition confidence calculations of the recognition engine, but the speech recognition engine may output text words which are input to the application program as text.
- the ASR system includes an acoustic front end (AFE) for receiving an audio signal from an input device, removing noise, identifying the parts of the audio signal containing speech, and parameterizing sections of the identified speech.
- a speech recognition engine receives inputs from the AFE and compares the received inputs with an acoustic model with reference to an application grammar (a database of significant words) and a baseform pool (a database of phones comprising known ways of speaking words or word sequences).
- the recognition engine outputs results comprising candidate matching paths between states of the acoustic model, a word or words from the application grammar corresponding to each candidate path, and confidence scores calculated for phones and paths.
- a further aspect of the invention comprises a method for calculating the confusability between sounds within the acoustic model of the ASR system.
- the confusability between two phones is based on a distance between the respective probability distributions associated with sounds within the acoustic model. A smaller distance is interpreted as meaning that two sounds will be more readily confused than sounds for which the distance is greater.
- the calculation of inherent confusability may be based on, for example, a Euclidean distance between the means of the probability distributions of any two sounds within the acoustic model taking into account the variance of the distributions.
- a calculated inherent confusability such as described above can be combined with dynamically calculated recognition confidence scores (based, for example, on closeness of an incoming acoustic segment to the mean of the probability distribution for a sound within the acoustic model).
- the results can be used when interrogating an application grammar to determine whether there are confusable words in the grammar which are relevant to a set of candidate matching phones.
- the combination of inherent confusability and dynamically calculated confidence scores can provide an improved evaluation of recognition confidence and improve selection of user prompts to enhance recognition.
- Embodiments of the invention may be used to modify the words in an application grammar to avoid possible confusions between items in the grammar. Calculations of inherent confusability between sounds in an acoustic model can be multipled by runtime recognition confidence scores to identify recognition difficulties, and then problematic words in the grammar (those that are highly confusable in the application environment) can be replaced with synonyms that are not as likely to be confused.
- Such program code may be made available as a set of one or more computer program products or modules comprising program code recorded on a recording medium.
- FIG. 1 is a schematic representation of an automatic speech recognition system, such as is known in the art
- FIGS. 2A and 2B represent paths through states and state transitions of an acoustic model
- FIG. 3 is a schematic representation of an automatic speech recognition system according to an embodiment of the invention.
- FIGS. 4 and 5 show the steps of two methods according to example embodiments of the invention.
- FIG. 6 shows an extract from an example conversation between a caller and an automatic speech recognition system according to an embodiment of the invention.
- An automatic speech recognition (ASR) system as shown in FIG. 1 comprises an input device 100 , such as a conventional microphone or telephone handset, an acoustic front end (AFE) 101 receiving inputs from the input device, a speech recognition engine 102 receiving inputs from the AFE, and a speech-responsive application program 103 that interfaces to the speech recognition engine.
- the application 103 defines a set of logical steps to be executed as part of the interaction between the user and the ASR system.
- the application 103 typically identifies to the user what input is required via user prompts.
- the user prompts may be text strings displayed on a screen or audio segments played to the user.
- the speech-responsive application uses the results of the speech recognition engine to perform actions based on the inputs.
- a user must input their account number and the ASR system is required to retrieve the relevant information (the account balance) from a database.
- the sequence of events within the interaction between the caller and ASR system may be represented as follows:
- the ASR system may have pre-recorded audio segments or may use text-to-speech synthesis.
- the machine uses a recognition engine, with grammars comprising possible input numbers and yes/no confirmation words. These grammars contain all the possible inputs that were envisaged by the application developer as likely responses to the prompts. The numbers grammar in this example will therefore contain all the digits from 0 to 9 inclusive, to allow the user to say any of those digits in any order.
- the yes/no confirmation grammar will typically contain not only “yes” and “no”, but also variants such as “yep”, “yeah”, “OK”, “correct”, “right”, “nah”, “nope”, and so on, together with optional items such as “please” and “thank you”.
- the input device 100 is responsible for capturing an audio signal and converting the signal to a digital signal for machine processing.
- the input device 100 may be remote from the other components of the ASR system.
- the digitised signal received from the input device 100 is presented to the acoustic front end (AFE) 101 of the ASR system.
- the AFE is responsible for:
- the AFE extracts a section of the signal (for example, a 100 msec time slice), applies a Hamming window and generates a smoothed spectral representation (an array of numbers defining a polynomial representation of the section). This array of numbers is passed to the recognition process of the recognition engine, which processes the array according to the acoustic model (AM). The AFE then returns to extract the next, possibly overlapping, section from the signal and repeats the described steps until the complete audio signal has been processed.
- a section of the signal for example, a 100 msec time slice
- a Hamming window an array of numbers defining a polynomial representation of the section.
- This array of numbers is passed to the recognition process of the recognition engine, which processes the array according to the acoustic model (AM).
- the AFE then returns to extract the next, possibly overlapping, section from the signal and repeats the described steps until the complete audio signal has been processed.
- the recognition engine 102 receives inputs from the acoustic front end (AFE) 101 and processes the inputs—applying an acoustic model (AM) with reference to an application grammar 105 and an associated baseform pool 104 .
- the application grammar 105 is typically an application-specific ‘mini-dictionary’ and ‘phrase book’ and includes words that the application developer thought users may use as inputs to the speech-enabled application.
- the application grammar could be a complete dictionary of a language, including common variations, but most applications require real-time responses and so limited-size application grammars are currently the norm.
- the baseform pool 105 comprises a pronunciation dictionary in the form of a repository of known pronunciations of phones (small identifiable units of sound) that may be included in an input word or phrase within the application grammar.
- the recognition engine attempts to match sounds within the input audio signal against phones within the acoustic model, to identify a likely ‘path’ between phones (that is, a sequence of states and state transitions within the acoustic model) corresponding to the sounds of the input audio signal.
- a likely ‘path’ between phones that is, a sequence of states and state transitions within the acoustic model
- Each phone is mapped to a set of states and transitions between states of the acoustic model, and each path comprises one or more phones.
- the acoustic model is typically a Hidden Markov Model (HMM), although other techniques are known.
- FIG. 2 is a simplified representation of some paths within a Hidden Markov Model. Each sound in the language can be represented as a path between states of the model, although there may be several paths that are each candidates for representing the same sound.
- Each state is a multi-dimensional Gaussian probability distribution, where the number of dimensions is fixed and relates to the number of terms used to describe the polynomial from the AFE processing. Transitions between states (including a “zero transition” to the same state) are also Gaussian probability distributions which refer to the likelihood that the current state could have been reached from the previous state.
- the recognition processing is described in more detail below, with reference to FIGS. 2A and 2B .
- the simplified path shown in FIG. 2A represents the beginning, middle and end of a single sound, say “Y” 201 as in the word “yes”.
- successive sections from the AFE 101 are received at the recognition engine 102 , they are compared to each of the states 202 with a view to identifying the state that is the most likely matching state for the respective section. If a good match is found at the current state, then the zero transition probability is queried. If a different state is a good match, then the transition probability between the current and next states is queried.
- a “path” 204 is traversed through the HMM. As the path progressively passes through states 202 and transitions 203 , a cumulative probability is calculated for the path.
- FIG. 2B is a simplified representation of moving from one phone label to the next, via two-alternative paths.
- a successful path which relates to the phone “Y” 205 in “yes” competes with the alternative path for the phone “J” 206 in “jelly”, because the input audio signal could be matched successfully to either of the alternatives.
- These two possibilities, “Y” and “J”, are both retained for now.
- the recognition engine 102 uses two additional inputs to process the audio signal: the application-specific grammar 105 and the pronunciation dictionary 104 referred to herein as the baseform pool 104 .
- the AM contains states and transitions which relate to parts (such as beginning, middle and end) of all sounds for a given natural language.
- the Gaussian distributions are ‘trained’ on audio data covering all possible sounds in all possible combinations for as many speakers as possible.
- This training step may include collating and summarising, the observed variations of the phone “Y” across all speakers who took part in the training set, and across all contexts represented in the training set.
- the recognition engine can recognise any valid sequence of sounds for that language. Because this could lead to a proliferation of paths, an application specific grammar and associated baseform pool are defined to speed and support recognition. Entries in the grammar are used to identify and therefore limit the words that will be recognised.
- the use of domain and application-specific grammars is common, although it is possible to have a general and complex grammar for the natural language.
- the grammar as defined earlier contains all the allowable words and phrases for a given application (i.e. words and phrases that the application will treat as meaningful inputs). For a banking application (such as the example given previously), the allowable words and phrases may be limited to digits and yes/no variants and so the application grammar does not contain other words such as proper names or literary terms.
- a baseform pool is generated during compilation which represents all possible pronunciations for the words in the grammar. These pronunciations—the baseforms—are the set of valid sound sequences that may be matched.
- the entries in the grammar comprise the set of words that will be recognised. For example, if the grammar contains only the words “Yes”, “No” and “Correct”, then only those words can produce a positive result. Entries in the baseform pool define the paths that can potentially be traversed in the acoustic model. For each word in the grammar, paths are retrieved from the baseform pool to constrain the number of paths which will be maintained from the matching step. Therefore, the grammar and baseform pool are used cooperatively to limit the overall search space within the AM. At runtime, they are used to restrict and/or “prune” paths through states of the AM.
- a single result is returned as the input audio is recognised as including the path Y-EH-S with a confidence score derived from the cumulative probability calculated for the path (for example a probability of 0.60, referred to herein as a confidence score of 60%).
- the single result includes the orthographically correct version ‘Yes’, as well as the baseforms Y, EH, S arranged in order, and the confidence score.
- an N-best list may be returned such as in the following example (which is merely illustrative):
- the above list includes the best and N-1 closest matches to sounds extracted from the input audio signal, which are found by applying the AM using sounds and words defined in the baseform pool and grammar.
- the baseforms, the orthographically correct version of the respective matching word and the associated confidence scores are returned within each result.
- the recognition engine calculates a confidence score associated with each identified candidate path—based on the closeness of the acoustic match and possibly also the speech context.
- the recognition engine either identifies a single word in the application grammar corresponding to the path having the highest confidence score among the identified paths, or records the confidence scores for a set of paths. Regardless of whether a single result or multiple results (a so-called ‘N-best list’) are returned, the application can use the confidence scores to determine whether or not to accept the result. For example, a high confidence score may be considered to justify accepting the result without a separate confirmation step, and a low score may be responded to by looping back and re-requesting input from the user or requesting confirmation of an identified match.
- ASR system in which recognition confidence scores are calculated by the recognition engine 102 for candidate matching phones of a first user input.
- the recognition confidence scores are used to control selection of a subsequent user prompt, which is associated with a subsequent step within the user's interaction with the ASR program.
- the invention also enables updating of an application grammar and an acoustic model as a consequence of replacing a user prompt that resulted in low confidence recognition scores. For example, if ‘Yes’ is associated with a low confidence score, the word ‘Yes’ may be replaced by ‘Yes please’ and the gammar modified accordingly.
- the calculated recognition confidence scores can show which input sounds were not recognized with high confidence. Different embodiments of the present invention make use of this information in different ways, as described below.
- the recognition engine uses phone-specific recognition confidence scores to identify phones that appear to require additional recognition training.
- the ASR program When generating a user prompt for a subsequent user interaction step with the speech-responsive application, the ASR program generates a prompt that invites input of words containing the phones for which additional training was determined to be required.
- the ASR program generates a prompt that is expected to avoid or reduce potential confusion by encouraging input of words other than those identified as low confidence matches.
- the application grammar may be dynamically modified in response to the selection or customization of user prompts, as explained below, although the application developer may include potential replacement words within the grammar from the outset.
- FIG. 3 shows the components of an automatic speech recognition (ASR) system according to an embodiment of the invention.
- the ASR system of FIG. 3 includes an acoustic front end (AFE) 101 and a recognition engine (RE) 102 referring to the application grammar 105 and baseform pool 104 .
- the ASR system of FIG. 3 also has additional components not found in the known ASR system of FIG. 1 , including a modified prompt database 106 and other components described below.
- Known ASR systems store audio files or text that are used to present results to a user or to indicate that input is required.
- the prompt database 106 also contains indicators of potential confusability and how they should be handled, as described below.
- the prompt database 106 is accessed by the application 103 to select user prompts automatically (selecting complete prompts or words for inclusion within user prompts). Use of the prompt database 106 is described in more detail below with reference to FIGS. 4 and 5 .
- part of a Dialogue Manager's job is to select appropriate prompts and the present invention could be implemented such that the recognition engine 102 selects user prompts.
- the grammar 105 is coded by the developer and the baseform pool 104 is derived from the specific application grammar. The process of deriving the baseforms or pronunciations is dependent on processing the results from training data.
- a homophony table 107 (a table of words that are spelt differently from each other but are pronounced the same and are therefore likely to be confused) is also generated as described below:
- the ASR system described here defers the step of highlighting likely confusions until speech is analyzed at runtime, so that the inherent confusability between words in the application grammar can be correlated with runtime recognition confidence scores.
- the components and runtime operations of the ASR system are described in more detail below with reference to FIGS. 3 , 4 and 5 .
- the sequence of method steps shown in FIG. 4 begins with receipt 300 of pre-processed, parameterized segments of audio data by the recognition engine 102 .
- the recognition engine 102 compares 301 the input audio segments against the acoustic model to identify candidate matching phones (sequences of states within the model) and sequences of candidate phones (referred to as ‘paths’ within the acoustic model) that are possible representations of the sounds spoken by a user.
- the recognition engine compares 302 the candidate paths with the application grammar 105 in an attempt to identify words in the grammar that match the sounds spoken by the user.
- the recognition engine 102 uses the closeness of input audio to the mean of the probability distributions associated with individual phones of the acoustic model for each candidate path to calculate 303 confidence scores for each candidate matching phone and word.
- the confidence scores are collected 303 in a phone confidence table 108 .
- the phone confidence table 108 contains the phones recognised, the associated confidence scores, the inherent confusability measure for each phone, and a cumulative confusability index for each word recognised. These will be used to evaluate the result as described below.
- the trend of these measures for individual phones is stored in a confidence trend database, also explained below but not shown in the figures.
- the confusability indices calculated separately and the runtime values for the specific application and specific user are all retained—so that the generalised confusion measures can be modified dynamically to the real deployment environment.
- the recognition confidence score calculated at runtime is multiplied with the inherent phone confusability indices calculated in advance. In this way, the confidence value will reduce in proportion to the inherent (i.e. language-general) and cumulative (ie. application-specific) indices if the latter two values indicate a high degree of confusability.
- a per-speaker As described above, three values are calculated and used when evaluating recognition confidence and likely confusions: a per-speaker, environment-specific run-time recognition confidence; an inherent per-phone confusability; and a cumulative confusability index for a word.
- the confidence score represents how the recognition engine is performing under the current conditions for the current speaker.
- the inherent per-phone confusability index represents how distinct the sound or sounds are that are being pronounced (in the particular language).
- the cumulative confusability index represents how important inherent confusability is for this specific application. The system is optimized if, all else being equal, the cumulative confusability index is as high as possible.
- the runtime confusion measure that combines the inherent confusability values and the run-time confidence is used by the speech-responsive application in one of the ways described below.
- user input prompts are adapted to try to elicit user inputs that include phones that were previously recognized with low recognition confidence.
- a number of optional user prompts are stored in a prompt database 106 and are associated with an application program.
- the optional prompts are each selectable by a prompt selector 110 , implemented within the recognition engine in this example embodiment, based on received user inputs and calculated recognition confidence values.
- Stored together with at least some user prompts are pointers to phones within the baseform pool 104 , which phones are expected to be included within the words spoken by a user in response to the respective prompt.
- the application program responds to certain user inputs by invoking 310 a prompt selector process 110 in response to results output by the recognition engine 102 (or the prompt selector may be implemented within the recognition engine 102 ).
- the prompt selector refers to the prompt database 106 to identify 311 a set of optional user prompts from which to select the next user prompt.
- the prompt selector also refers to the phone confidence table 108 to identify phones previously recognised with low confidence that are relevant to the set of optional user prompts, and refers to the homophony table to identify any words that could cause confusion issues.
- the prompt selector selects 312 a user prompt that is expected to encourage input of one or more phones previously recognized with low confidence.
- the selected prompt is then returned to the application.
- the input can now be used in one of two ways: either used directly in a recognition engine which supports adaptation to provide real-time training data, or separately recorded with the associated recognition result for offline processing. In either case, the expectation is that over time the additional training data would help increase recognition success and therefore the runtime confidence measure would increase.
- the targetted recognition training is most effective if a user is prompted to input a word or phrase that includes at least one high recognition confidence phone and the low recognition confidence phone or phones that require training. Any additional prompted inputs (such as prompted inputs of ‘Yes please’ instead of ‘Yes’) will aid recognition training because there are more data to analyze.
- the selection from a set of optional user prompts may comprise selecting a word or phrase that the user is invited to say in response to a question, and the selected word or phrase is used to generate a user-specific customized dialogue.
- a standard question asked by an ASR system may be “If you wish to know opening times, say ⁇ PHRASE>”, and the optional user prompts held in the prompt database may be phrases such as ‘Yes’, ‘Yes please’, ‘Correct’, ‘Opening times’, etc.
- the Prompt selector may select ‘Yes please’ in response to a previous recognition of ‘Yes’ with a low recognition score, since this will allow targetted recognition training for the phones within the word ‘Yes’. Because both the sounds which previously produced high recognition confidence are known, it is more likely that the recognition engine will return a successful result to the current application. Secondly, because both those sounds and the ones which produce low scores are known, it is easier to identify which parts of the audio input are needed for training purposes.
- a synonym dictionary 109 is queried 320 to find possible synonyms for the word or phrase which is confusable.
- the synonym dictionary contains words for replacing a word that includes a low confidence phone, which replacement word can be added to the grammar 105 if not already included and can be weighted to be more likely than the original term.
- the replacement words are combined 330 with an associated prompt text or audio file to generate a user prompt that is expected to encourage input of the replacement term.
- a replacement synonym is added to a prompt text or audio file to generate a modified user prompt.
- the prompt database 106 can be updated if required.
- An original application prompt includes the words:
- the confidence trend database mentioned above is also updated.
- This confidence trend database can be periodically queried to highlight phones in whatever context which are leading to lower recognition success rates.
- dynamic adaptation can be triggered in the AM.
- Adaptation involves capturing runtime acoustic data and not only processing as normal (see FIG. 1 and the description above), but additionally weighting the Gaussian distributions (‘retraining’) using the runtime acoustic data. This can achieve improved matching over time, but the process is ‘blind’ in the sense that it is used irrespective of where issues may arise in sections of the HMM.
- actual performance information is captured and this could trigger adaptation in a more targetted manner.
- the performance information could be used in one embodiment to generate a separate retraining report for the system administrator to review and address as required.
- FIG. 6 shows interactions between a caller and an ASR system.
- the ASR system includes a prompt selector and a user prompt database storing replacement user prompts.
- a timeline is shown on the left of FIG. 6 .
- the interaction begins with a caller ringing a telephone number of the ASR system to determine information relating to films showing at a cinema.
- the ASR system responds to the caller by asking “Which cinema please?”.
- the caller responds by identifying a cinema, for example “The Odeon in Southampton”.
- the ASR system analyzes the spoken input to identify possible matching phones and words, and to determine recognition confidence scores from the cumulative probability for matched paths. Where recognition confidence scores are lower than a preset threshold, this recognition difficulty is responded to by selecting an alternative word or phrase from a prompt database.
- a first ASR system if an input word or phrase is recognized with low confidence, subsequent generation of user prompts involves selecting a user prompt from a set of alternative user prompts with the selection being based on which user prompts are associated with phones identified as having low confidence recognition.
- the ASR system selects user prompts that are associated with low confidence, in order to make use of any additional opportunities for speech recognition training.
- a subsequent user prompt which is likely to be responded to with the same input word or phrase is not used but is replaced by a user prompt that does not invite the same response.
- the replacement user prompt is selected in the expectation that the spoken response to this replacement user prompt will be recognizable with greater confidence than the problematic input speech.
- a customized dialogue is dynamically generated for the particular caller at a particular time and in a particular environment using the replacement prompt.
- replacement prompts invite a very specific response from callers (such as “Thank you. To listen to information on ⁇ X>, say ‘YES YOUR’”).
- the subsequent response from the caller may also be analyzed, a confidence score calculated and if necessary the confidence score may be used to influence subsequent generation of user prompts.
- the steps of adjusting subsequent user prompts in response to confidence scores associated with the recognition of user inputs can be repeated as many times as required, such as until the speech recognition program achieves a sufficient recognition confidence to permit more free-form inputs, or until the end of the conversation.
- An ASR system may perform functions that would otherwise require many personnel to handle incoming calls, including answering trivial questions that can be readily automated.
- An ASR system may implement menu selection operations to direct a caller to recorded information or to the correct person or department within an organisation.
- confidence scores can show which areas of a language an automatic speech recognition system has most difficulty recognizing.
- Recognition problems and the evaluation of recognition confidence scores may be caller-specific, since certain user prompts may result in low confidence recognition for some specific callers only, or may be environment-specific in the sense that certain user inputs may be more prone to distortion in background noise, for example.
- user prompts are adjusted to encourage the caller to use words that are expected to be recognized with higher confidence scores, or to provide a training opportunity within the context of a functional dialogue between the caller and the speech recognition system.
- Recognition confidence can then be improved with minimal inconvenience to callers, because the adjustment of user prompts and any retraining can be achieved without a dedicated recognition training period and potentially without requiring callers to repeat their inputs. Callers typically require that their telephone calls achieve a desired result very quickly, especially when speaking to an automatic speech recognition system.
- a recognition grammar may be modified on the basis of performance, and the recognition engine's acoustic model may be optimised to improve future performance.
- the prompt selector was described as a component of the recognition engine, with the application interfacing to the recognition engine to invoke functions of the prompt selector.
- the prompt selector is implemented within a speech-responsive application program but refers to the recognition confidence table populated by the recognition engine.
- the recognition engine and the prompt database can interface to a conventional speech-recognition application and the user prompt selection operations are performed transparently from the perspective of the application.
Landscapes
- Engineering & Computer Science (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Human Computer Interaction (AREA)
- Physics & Mathematics (AREA)
- Acoustics & Sound (AREA)
- Multimedia (AREA)
- Machine Translation (AREA)
- Telephonic Communication Services (AREA)
Abstract
Description
- 1. A caller accesses an ASR system to request a service;
- 2. The ASR system prompts the caller to input an account number (“Please Say your account number”);
- 3. The caller speaks the account number (“12345”);
- 4. The ASR system takes the audio input from the previous step,and uses a recognition engine to convert to text;
- 5. The ASR system echoes back the recognised account number and seeks confirmation (“You said 12345. If this is correct, please say Yes. If this is not correct, please say No.”);
- 6. The caller confirms (“Yes” in this case);
- 7. The ASR system accesses a database using the account number 12345 and retrieves the account balance;
- 8. The ASR system outputs the retrieved information to the caller (“The account balance for account 12345 is £100 credit”).
- 1. pre-processing the audio signal to remove noise;
- 2. establishing which part of the audio signal contains valid speech; and
- 3. parameterizing successive sections of the signal to be matched by the recognizer.
Y-EH-S | 60% | Yes | ||
Y-EH-L-OW | 20% | Yellow | ||
Y-OH-TD | 15% | Yacht | ||
J-EH-L-IY | 3% | Jelly | ||
N-OW | 2% | No | ||
-
- 1. based on the raw AM, a Euclidean distance measure is calculated for each and every phone by comparison to each and every other phone. Thus, for Y, there will be a data array including a multiplicity of entries that show how close in acoustic terms Y is to every other phone. These are the raw confusability indices: a high score (greater distance) would mean the two phones are unlikely to be confused (for example ‘O’ and ‘B’); a low score (smaller distance) would mean that the two phones are acoustically similar and therefore potentially confusable (for example ‘P’ and ‘B’, or ‘S’ and ‘F’).
- The following illustrates a first exemplary method for calculating the Euclidean distance between states x of two different sounds z and y. Let xazl represent the mean value of the first parameter of state a for sound z, and Xayl represent the mean for the first parameter of state a sound y. The Euclidean distance between them is defined as: √{square root over ((xazl−xayl)2)}. For a multidimensional Gaussian distribution as in the HMM network of a
typical recognition engine 102, this calculation is performed for x from 1 to M, where M represents the number of states used to represent a given sound. The Euclidean distance per state is now given by: √{square root over (Σ(xazc−xayc)2)} where c is from 1 to N, where N represents the number of states used. - Knowing the variance around the mean, and the orientation of a vector between the two means, allows an extension and better confusion indicator. Extending the approach approve, the preferred distance measure is given by: :√{square root over (Σ((xazc±σazc)−(xayc±σ2 ayc))2)}{square root over (Σ((xazc±σazc)−(xayc±σ2 ayc))2)}
- 2. in addition, a ‘phone confusability index’ is calculated for each phone, for example as the average of the confusability indices below a given threshold for each phone of a phone path. This phone confusability index is referred to herein as the inherent confusability measure for the phone.
- 3. taking the application-specific baseform pool (a database of phone paths relevant to the application grammar), the raw confusability indices are used to calculate, for each word of the application grammar, a cumulative confusability index representing how confusable two or more words are likely to be. The cumulative confusability index for pairs of words is stored in the homophony table 107.
-
- let (a) be the confidence score for a specific recognition attempt (for example 70%); and
- let (b) be the inherent confusability for all the sounds in the result by comparison to the other sounds in the acoustic model (for example 25%, if the sounds in the result are highly confusable).
- Additionally, let (c) be the cumulative confusability for the specific word or words in the result, by comparison with all other words in the grammar (for example a value of 80% if we predict that this word or these words are not easily confused with the other words in the grammar, even though individual sounds might be confusable).
- Values (b) and (c) have been calculated offline. Multiplying them together, we get 20%.
- Multiplying the confidence score (a) with this 20% result, gives a runtime confidence measure of 14%.
Claims (23)
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US14/191,176 US9502024B2 (en) | 2004-12-01 | 2014-02-26 | Methods, apparatus and computer programs for automatic speech recognition |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
GB0426347.1 | 2004-12-01 | ||
GBGB0426347.1A GB0426347D0 (en) | 2004-12-01 | 2004-12-01 | Methods, apparatus and computer programs for automatic speech recognition |
Related Child Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US14/191,176 Continuation US9502024B2 (en) | 2004-12-01 | 2014-02-26 | Methods, apparatus and computer programs for automatic speech recognition |
Publications (2)
Publication Number | Publication Date |
---|---|
US20060116877A1 US20060116877A1 (en) | 2006-06-01 |
US8694316B2 true US8694316B2 (en) | 2014-04-08 |
Family
ID=34043856
Family Applications (2)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US11/255,201 Active 2031-04-10 US8694316B2 (en) | 2004-12-01 | 2005-10-20 | Methods, apparatus and computer programs for automatic speech recognition |
US14/191,176 Active US9502024B2 (en) | 2004-12-01 | 2014-02-26 | Methods, apparatus and computer programs for automatic speech recognition |
Family Applications After (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US14/191,176 Active US9502024B2 (en) | 2004-12-01 | 2014-02-26 | Methods, apparatus and computer programs for automatic speech recognition |
Country Status (3)
Country | Link |
---|---|
US (2) | US8694316B2 (en) |
CN (1) | CN1783213B (en) |
GB (1) | GB0426347D0 (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US9502024B2 (en) | 2004-12-01 | 2016-11-22 | Nuance Communications, Inc. | Methods, apparatus and computer programs for automatic speech recognition |
US9953646B2 (en) | 2014-09-02 | 2018-04-24 | Belleau Technologies | Method and system for dynamic speech recognition and tracking of prewritten script |
US10102771B2 (en) | 2013-04-26 | 2018-10-16 | Wistron Corporation | Method and device for learning language and computer readable recording medium |
Families Citing this family (232)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8645137B2 (en) | 2000-03-16 | 2014-02-04 | Apple Inc. | Fast, language-independent method for user authentication by voice |
WO2006093092A1 (en) * | 2005-02-28 | 2006-09-08 | Honda Motor Co., Ltd. | Conversation system and conversation software |
US7983914B2 (en) * | 2005-08-10 | 2011-07-19 | Nuance Communications, Inc. | Method and system for improved speech recognition by degrading utterance pronunciations |
US8548799B2 (en) * | 2005-08-10 | 2013-10-01 | Microsoft Corporation | Methods and apparatus to help users of a natural language system formulate queries |
US8677377B2 (en) | 2005-09-08 | 2014-03-18 | Apple Inc. | Method and apparatus for building an intelligent automated assistant |
US20070132834A1 (en) * | 2005-12-08 | 2007-06-14 | International Business Machines Corporation | Speech disambiguation in a composite services enablement environment |
US8811369B2 (en) | 2006-01-11 | 2014-08-19 | Qualcomm Incorporated | Methods and apparatus for supporting multiple communications modes of operation |
US8755362B2 (en) | 2006-01-11 | 2014-06-17 | Qualcomm Incorporated | Wireless communication methods and apparatus supporting paging and peer to peer communications |
US7562815B2 (en) * | 2006-06-01 | 2009-07-21 | Card Logistics Properties, Ltd. | Gaming smart card |
CN101118745B (en) * | 2006-08-04 | 2011-01-19 | 中国科学院声学研究所 | Confidence degree quick acquiring method in speech identification system |
US9318108B2 (en) | 2010-01-18 | 2016-04-19 | Apple Inc. | Intelligent automated assistant |
KR100897554B1 (en) * | 2007-02-21 | 2009-05-15 | 삼성전자주식회사 | Distributed speech recognition sytem and method and terminal for distributed speech recognition |
US8977255B2 (en) | 2007-04-03 | 2015-03-10 | Apple Inc. | Method and system for operating a multi-function portable electronic device using voice-activation |
US9330720B2 (en) | 2008-01-03 | 2016-05-03 | Apple Inc. | Methods and apparatus for altering audio output signals |
US8996376B2 (en) | 2008-04-05 | 2015-03-31 | Apple Inc. | Intelligent text-to-speech conversion |
US8595501B2 (en) | 2008-05-09 | 2013-11-26 | Qualcomm Incorporated | Network helper for authentication between a token and verifiers |
US10496753B2 (en) | 2010-01-18 | 2019-12-03 | Apple Inc. | Automatically adapting user interfaces for hands-free interaction |
US8566076B2 (en) * | 2008-05-28 | 2013-10-22 | International Business Machines Corporation | System and method for applying bridging models for robust and efficient speech to speech translation |
US20100030549A1 (en) | 2008-07-31 | 2010-02-04 | Lee Michael M | Mobile device having human language translation capability with positional feedback |
US20100070932A1 (en) * | 2008-09-18 | 2010-03-18 | Nissan Technical Center North America, Inc. | Vehicle on-board device |
US8676904B2 (en) | 2008-10-02 | 2014-03-18 | Apple Inc. | Electronic devices with voice command and contextual data processing capabilities |
US8155961B2 (en) * | 2008-12-09 | 2012-04-10 | Nokia Corporation | Adaptation of automatic speech recognition acoustic models |
WO2010067118A1 (en) * | 2008-12-11 | 2010-06-17 | Novauris Technologies Limited | Speech recognition involving a mobile device |
US20120004910A1 (en) * | 2009-05-07 | 2012-01-05 | Romulo De Guzman Quidilig | System and method for speech processing and speech to text |
US10241644B2 (en) | 2011-06-03 | 2019-03-26 | Apple Inc. | Actionable reminder entries |
US10241752B2 (en) | 2011-09-30 | 2019-03-26 | Apple Inc. | Interface for a virtual digital assistant |
US20120309363A1 (en) | 2011-06-03 | 2012-12-06 | Apple Inc. | Triggering notifications associated with tasks items that represent tasks to perform |
US9858925B2 (en) | 2009-06-05 | 2018-01-02 | Apple Inc. | Using context information to facilitate processing of commands in a virtual assistant |
US9431006B2 (en) | 2009-07-02 | 2016-08-30 | Apple Inc. | Methods and apparatuses for automatic speech recognition |
US9653066B2 (en) * | 2009-10-23 | 2017-05-16 | Nuance Communications, Inc. | System and method for estimating the reliability of alternate speech recognition hypotheses in real time |
US10553209B2 (en) | 2010-01-18 | 2020-02-04 | Apple Inc. | Systems and methods for hands-free notification summaries |
US10679605B2 (en) | 2010-01-18 | 2020-06-09 | Apple Inc. | Hands-free list-reading by intelligent automated assistant |
US10705794B2 (en) | 2010-01-18 | 2020-07-07 | Apple Inc. | Automatically adapting user interfaces for hands-free interaction |
US10276170B2 (en) | 2010-01-18 | 2019-04-30 | Apple Inc. | Intelligent automated assistant |
JP5633042B2 (en) * | 2010-01-28 | 2014-12-03 | 本田技研工業株式会社 | Speech recognition apparatus, speech recognition method, and speech recognition robot |
US8682667B2 (en) | 2010-02-25 | 2014-03-25 | Apple Inc. | User profiling for selecting user specific voice input processing information |
CN102237087B (en) * | 2010-04-27 | 2014-01-01 | 中兴通讯股份有限公司 | Voice control method and voice control device |
US8738377B2 (en) * | 2010-06-07 | 2014-05-27 | Google Inc. | Predicting and learning carrier phrases for speech input |
US9031844B2 (en) | 2010-09-21 | 2015-05-12 | Microsoft Technology Licensing, Llc | Full-sequence training of deep structures for speech recognition |
US8639508B2 (en) * | 2011-02-14 | 2014-01-28 | General Motors Llc | User-specific confidence thresholds for speech recognition |
US9262612B2 (en) | 2011-03-21 | 2016-02-16 | Apple Inc. | Device access using voice authentication |
US10057736B2 (en) | 2011-06-03 | 2018-08-21 | Apple Inc. | Active transport based notifications |
WO2013033119A1 (en) * | 2011-08-29 | 2013-03-07 | Accumente, Llc | Utilizing multiple processing units for rapid training of hidden markov models |
US8994660B2 (en) | 2011-08-29 | 2015-03-31 | Apple Inc. | Text correction processing |
US9002703B1 (en) * | 2011-09-28 | 2015-04-07 | Amazon Technologies, Inc. | Community audio narration generation |
US9235799B2 (en) | 2011-11-26 | 2016-01-12 | Microsoft Technology Licensing, Llc | Discriminative pretraining of deep neural networks |
US8594281B2 (en) * | 2012-01-10 | 2013-11-26 | Bank Of America Corporation | Dynamic menu framework |
US10134385B2 (en) | 2012-03-02 | 2018-11-20 | Apple Inc. | Systems and methods for name pronunciation |
US9483461B2 (en) | 2012-03-06 | 2016-11-01 | Apple Inc. | Handling speech synthesis of content for multiple languages |
US9280610B2 (en) | 2012-05-14 | 2016-03-08 | Apple Inc. | Crowd sourcing information to fulfill user requests |
US10417037B2 (en) | 2012-05-15 | 2019-09-17 | Apple Inc. | Systems and methods for integrating third party services with a digital assistant |
US9721563B2 (en) | 2012-06-08 | 2017-08-01 | Apple Inc. | Name recognition system |
US9734839B1 (en) * | 2012-06-20 | 2017-08-15 | Amazon Technologies, Inc. | Routing natural language commands to the appropriate applications |
US9495129B2 (en) | 2012-06-29 | 2016-11-15 | Apple Inc. | Device, method, and user interface for voice-activated navigation and browsing of a document |
US9536528B2 (en) * | 2012-07-03 | 2017-01-03 | Google Inc. | Determining hotword suitability |
CN102760433A (en) * | 2012-07-06 | 2012-10-31 | 广东美的制冷设备有限公司 | Sound control remote controller and control method of networked household appliances |
US9576574B2 (en) | 2012-09-10 | 2017-02-21 | Apple Inc. | Context-sensitive handling of interruptions by intelligent digital assistant |
US8700396B1 (en) * | 2012-09-11 | 2014-04-15 | Google Inc. | Generating speech data collection prompts |
US9646604B2 (en) * | 2012-09-15 | 2017-05-09 | Avaya Inc. | System and method for dynamic ASR based on social media |
US9547647B2 (en) | 2012-09-19 | 2017-01-17 | Apple Inc. | Voice-based media searching |
US9601111B2 (en) * | 2012-11-13 | 2017-03-21 | GM Global Technology Operations LLC | Methods and systems for adapting speech systems |
US9477925B2 (en) | 2012-11-20 | 2016-10-25 | Microsoft Technology Licensing, Llc | Deep neural networks training for speech and pattern recognition |
US8977555B2 (en) * | 2012-12-20 | 2015-03-10 | Amazon Technologies, Inc. | Identification of utterance subjects |
CN103077714B (en) * | 2013-01-29 | 2015-07-08 | 华为终端有限公司 | Information identification method and apparatus |
CN104969289B (en) | 2013-02-07 | 2021-05-28 | 苹果公司 | Voice trigger of digital assistant |
CN103198829A (en) * | 2013-02-25 | 2013-07-10 | 惠州市车仆电子科技有限公司 | Method, device and equipment of reducing interior noise and improving voice recognition rate |
US9368114B2 (en) | 2013-03-14 | 2016-06-14 | Apple Inc. | Context-sensitive handling of interruptions |
US10652394B2 (en) | 2013-03-14 | 2020-05-12 | Apple Inc. | System and method for processing voicemail |
US10748529B1 (en) | 2013-03-15 | 2020-08-18 | Apple Inc. | Voice activated device for use with a voice-based digital assistant |
WO2014144579A1 (en) | 2013-03-15 | 2014-09-18 | Apple Inc. | System and method for updating an adaptive speech recognition model |
US9922642B2 (en) | 2013-03-15 | 2018-03-20 | Apple Inc. | Training an at least partial voice command system |
WO2014197334A2 (en) | 2013-06-07 | 2014-12-11 | Apple Inc. | System and method for user-specified pronunciation of words for speech synthesis and recognition |
WO2014197336A1 (en) | 2013-06-07 | 2014-12-11 | Apple Inc. | System and method for detecting errors in interactions with a voice-based digital assistant |
US9582608B2 (en) | 2013-06-07 | 2017-02-28 | Apple Inc. | Unified ranking with entropy-weighted information for phrase-based semantic auto-completion |
CN104238379B (en) * | 2013-06-07 | 2017-07-28 | 艾默生过程控制流量技术有限公司 | Transmitter, field instrument and the method for controlling transmitter |
WO2014197335A1 (en) | 2013-06-08 | 2014-12-11 | Apple Inc. | Interpreting and acting upon commands that involve sharing information with remote devices |
EP3937002A1 (en) | 2013-06-09 | 2022-01-12 | Apple Inc. | Device, method, and graphical user interface for enabling conversation persistence across two or more instances of a digital assistant |
US10176167B2 (en) | 2013-06-09 | 2019-01-08 | Apple Inc. | System and method for inferring user intent from speech inputs |
CN105265005B (en) | 2013-06-13 | 2019-09-17 | 苹果公司 | System and method for the urgent call initiated by voice command |
US9280970B1 (en) * | 2013-06-25 | 2016-03-08 | Google Inc. | Lattice semantic parsing |
WO2015020942A1 (en) | 2013-08-06 | 2015-02-12 | Apple Inc. | Auto-activating smart responses based on activities from remote devices |
US9721570B1 (en) * | 2013-12-17 | 2017-08-01 | Amazon Technologies, Inc. | Outcome-oriented dialogs on a speech recognition platform |
US9817813B2 (en) * | 2014-01-08 | 2017-11-14 | Genesys Telecommunications Laboratories, Inc. | Generalized phrases in automatic speech recognition systems |
CN103903615B (en) * | 2014-03-10 | 2018-11-09 | 联想(北京)有限公司 | A kind of information processing method and electronic equipment |
CN109949815B (en) * | 2014-04-07 | 2024-06-07 | 三星电子株式会社 | Electronic device |
TW201543361A (en) * | 2014-05-07 | 2015-11-16 | Giga Byte Tech Co Ltd | Input system for rapidly activating macro function |
US9620105B2 (en) | 2014-05-15 | 2017-04-11 | Apple Inc. | Analyzing audio input for efficient speech and music recognition |
US10592095B2 (en) | 2014-05-23 | 2020-03-17 | Apple Inc. | Instantaneous speaking of content on touch devices |
US9502031B2 (en) | 2014-05-27 | 2016-11-22 | Apple Inc. | Method for supporting dynamic grammars in WFST-based ASR |
US9715875B2 (en) | 2014-05-30 | 2017-07-25 | Apple Inc. | Reducing the need for manual start/end-pointing and trigger phrases |
US9430463B2 (en) | 2014-05-30 | 2016-08-30 | Apple Inc. | Exemplar-based natural language processing |
US9734193B2 (en) | 2014-05-30 | 2017-08-15 | Apple Inc. | Determining domain salience ranking from ambiguous words in natural speech |
US9842101B2 (en) | 2014-05-30 | 2017-12-12 | Apple Inc. | Predictive conversion of language input |
US10289433B2 (en) | 2014-05-30 | 2019-05-14 | Apple Inc. | Domain specific language for encoding assistant dialog |
US9785630B2 (en) | 2014-05-30 | 2017-10-10 | Apple Inc. | Text prediction using combined word N-gram and unigram language models |
WO2015184186A1 (en) | 2014-05-30 | 2015-12-03 | Apple Inc. | Multi-command single utterance input method |
US10170123B2 (en) | 2014-05-30 | 2019-01-01 | Apple Inc. | Intelligent assistant for home automation |
US9760559B2 (en) | 2014-05-30 | 2017-09-12 | Apple Inc. | Predictive text input |
US10078631B2 (en) | 2014-05-30 | 2018-09-18 | Apple Inc. | Entropy-guided text prediction using combined word and character n-gram language models |
US9633004B2 (en) | 2014-05-30 | 2017-04-25 | Apple Inc. | Better resolution when referencing to concepts |
US10659851B2 (en) | 2014-06-30 | 2020-05-19 | Apple Inc. | Real-time digital assistant knowledge updates |
US9338493B2 (en) | 2014-06-30 | 2016-05-10 | Apple Inc. | Intelligent automated assistant for TV user interactions |
US10446141B2 (en) * | 2014-08-28 | 2019-10-15 | Apple Inc. | Automatic speech recognition based on user feedback |
US9818400B2 (en) | 2014-09-11 | 2017-11-14 | Apple Inc. | Method and apparatus for discovering trending terms in speech requests |
US10789041B2 (en) | 2014-09-12 | 2020-09-29 | Apple Inc. | Dynamic thresholds for always listening speech trigger |
US9668121B2 (en) | 2014-09-30 | 2017-05-30 | Apple Inc. | Social reminders |
US9886432B2 (en) | 2014-09-30 | 2018-02-06 | Apple Inc. | Parsimonious handling of word inflection via categorical stem + suffix N-gram language models |
US10127911B2 (en) | 2014-09-30 | 2018-11-13 | Apple Inc. | Speaker identification and unsupervised speaker adaptation techniques |
US10074360B2 (en) | 2014-09-30 | 2018-09-11 | Apple Inc. | Providing an indication of the suitability of speech recognition |
US9646609B2 (en) | 2014-09-30 | 2017-05-09 | Apple Inc. | Caching apparatus for serving phonetic pronunciations |
US10552013B2 (en) | 2014-12-02 | 2020-02-04 | Apple Inc. | Data detection |
US9711141B2 (en) | 2014-12-09 | 2017-07-18 | Apple Inc. | Disambiguating heteronyms in speech synthesis |
TWI590240B (en) * | 2014-12-30 | 2017-07-01 | 鴻海精密工業股份有限公司 | Meeting minutes device and method thereof for automatically creating meeting minutes |
TWI616868B (en) * | 2014-12-30 | 2018-03-01 | 鴻海精密工業股份有限公司 | Meeting minutes device and method thereof for automatically creating meeting minutes |
CN104572009B (en) * | 2015-01-28 | 2018-01-09 | 合肥联宝信息技术有限公司 | A kind of audio control method and device of adaptive external environment |
CN105989838B (en) * | 2015-01-30 | 2019-09-06 | 展讯通信(上海)有限公司 | Audio recognition method and device |
US10121466B2 (en) * | 2015-02-11 | 2018-11-06 | Hand Held Products, Inc. | Methods for training a speech recognition system |
US9865280B2 (en) | 2015-03-06 | 2018-01-09 | Apple Inc. | Structured dictation using intelligent automated assistants |
US9721566B2 (en) | 2015-03-08 | 2017-08-01 | Apple Inc. | Competing devices responding to voice triggers |
US10567477B2 (en) | 2015-03-08 | 2020-02-18 | Apple Inc. | Virtual assistant continuity |
US9886953B2 (en) | 2015-03-08 | 2018-02-06 | Apple Inc. | Virtual assistant activation |
US9899019B2 (en) | 2015-03-18 | 2018-02-20 | Apple Inc. | Systems and methods for structured stem and suffix language models |
US9842105B2 (en) | 2015-04-16 | 2017-12-12 | Apple Inc. | Parsimonious continuous-space phrase representations for natural language processing |
CN107112005A (en) * | 2015-04-17 | 2017-08-29 | 微软技术许可有限责任公司 | Depth nerve SVMs |
US10460227B2 (en) | 2015-05-15 | 2019-10-29 | Apple Inc. | Virtual assistant in a communication session |
US10200824B2 (en) | 2015-05-27 | 2019-02-05 | Apple Inc. | Systems and methods for proactively identifying and surfacing relevant content on a touch-sensitive device |
US10083688B2 (en) | 2015-05-27 | 2018-09-25 | Apple Inc. | Device voice control for selecting a displayed affordance |
US10127220B2 (en) | 2015-06-04 | 2018-11-13 | Apple Inc. | Language identification from short strings |
US10101822B2 (en) | 2015-06-05 | 2018-10-16 | Apple Inc. | Language input correction |
US9578173B2 (en) | 2015-06-05 | 2017-02-21 | Apple Inc. | Virtual assistant aided communication with 3rd party service in a communication session |
US10186254B2 (en) | 2015-06-07 | 2019-01-22 | Apple Inc. | Context-based endpoint detection |
US10255907B2 (en) | 2015-06-07 | 2019-04-09 | Apple Inc. | Automatic accent detection using acoustic models |
US11025565B2 (en) | 2015-06-07 | 2021-06-01 | Apple Inc. | Personalized prediction of responses for instant messaging |
US20160378747A1 (en) | 2015-06-29 | 2016-12-29 | Apple Inc. | Virtual assistant for media playback |
US10515150B2 (en) * | 2015-07-14 | 2019-12-24 | Genesys Telecommunications Laboratories, Inc. | Data driven speech enabled self-help systems and methods of operating thereof |
US10747498B2 (en) | 2015-09-08 | 2020-08-18 | Apple Inc. | Zero latency digital assistant |
US10740384B2 (en) | 2015-09-08 | 2020-08-11 | Apple Inc. | Intelligent automated assistant for media search and playback |
US10331312B2 (en) | 2015-09-08 | 2019-06-25 | Apple Inc. | Intelligent automated assistant in a media environment |
US10671428B2 (en) | 2015-09-08 | 2020-06-02 | Apple Inc. | Distributed personal assistant |
US9697820B2 (en) | 2015-09-24 | 2017-07-04 | Apple Inc. | Unit-selection text-to-speech synthesis using concatenation-sensitive neural networks |
US10366158B2 (en) | 2015-09-29 | 2019-07-30 | Apple Inc. | Efficient word encoding for recurrent neural network language models |
US11010550B2 (en) | 2015-09-29 | 2021-05-18 | Apple Inc. | Unified language modeling framework for word prediction, auto-completion and auto-correction |
US11587559B2 (en) | 2015-09-30 | 2023-02-21 | Apple Inc. | Intelligent device identification |
US10691473B2 (en) | 2015-11-06 | 2020-06-23 | Apple Inc. | Intelligent automated assistant in a messaging environment |
US10956666B2 (en) | 2015-11-09 | 2021-03-23 | Apple Inc. | Unconventional virtual assistant interactions |
US10049668B2 (en) | 2015-12-02 | 2018-08-14 | Apple Inc. | Applying neural network language models to weighted finite state transducers for automatic speech recognition |
CN105549949A (en) * | 2015-12-18 | 2016-05-04 | 合肥寰景信息技术有限公司 | Voice help prompt method of network community |
CN105549948A (en) * | 2015-12-18 | 2016-05-04 | 合肥寰景信息技术有限公司 | Voice operation method applied to network community |
CN105551489A (en) * | 2015-12-18 | 2016-05-04 | 合肥寰景信息技术有限公司 | Voice control platform of network community |
CN105653595A (en) * | 2015-12-18 | 2016-06-08 | 合肥寰景信息技术有限公司 | Intelligent voice assistance type network community |
US10223066B2 (en) | 2015-12-23 | 2019-03-05 | Apple Inc. | Proactive assistance based on dialog communication between devices |
CN106935239A (en) * | 2015-12-29 | 2017-07-07 | 阿里巴巴集团控股有限公司 | The construction method and device of a kind of pronunciation dictionary |
US10446143B2 (en) | 2016-03-14 | 2019-10-15 | Apple Inc. | Identification of voice inputs providing credentials |
JP6696803B2 (en) * | 2016-03-15 | 2020-05-20 | 本田技研工業株式会社 | Audio processing device and audio processing method |
US10152507B2 (en) | 2016-03-22 | 2018-12-11 | International Business Machines Corporation | Finding of a target document in a spoken language processing |
CN105872687A (en) * | 2016-03-31 | 2016-08-17 | 乐视控股(北京)有限公司 | Method and device for controlling intelligent equipment through voice |
CN105957535A (en) * | 2016-04-15 | 2016-09-21 | 青岛克路德机器人有限公司 | Robot voice signal detecting and identifying system |
US10192555B2 (en) * | 2016-04-28 | 2019-01-29 | Microsoft Technology Licensing, Llc | Dynamic speech recognition data evaluation |
US9934775B2 (en) | 2016-05-26 | 2018-04-03 | Apple Inc. | Unit-selection text-to-speech synthesis based on predicted concatenation parameters |
US9972304B2 (en) | 2016-06-03 | 2018-05-15 | Apple Inc. | Privacy preserving distributed evaluation framework for embedded personalized systems |
US10249300B2 (en) | 2016-06-06 | 2019-04-02 | Apple Inc. | Intelligent list reading |
US10049663B2 (en) | 2016-06-08 | 2018-08-14 | Apple, Inc. | Intelligent automated assistant for media exploration |
DK179309B1 (en) | 2016-06-09 | 2018-04-23 | Apple Inc | Intelligent automated assistant in a home environment |
US10192552B2 (en) | 2016-06-10 | 2019-01-29 | Apple Inc. | Digital assistant providing whispered speech |
US10586535B2 (en) | 2016-06-10 | 2020-03-10 | Apple Inc. | Intelligent digital assistant in a multi-tasking environment |
US10490187B2 (en) | 2016-06-10 | 2019-11-26 | Apple Inc. | Digital assistant providing automated status report |
US10509862B2 (en) | 2016-06-10 | 2019-12-17 | Apple Inc. | Dynamic phrase expansion of language input |
US10067938B2 (en) | 2016-06-10 | 2018-09-04 | Apple Inc. | Multilingual word prediction |
DK179415B1 (en) | 2016-06-11 | 2018-06-14 | Apple Inc | Intelligent device arbitration and control |
DK179343B1 (en) | 2016-06-11 | 2018-05-14 | Apple Inc | Intelligent task discovery |
DK179049B1 (en) | 2016-06-11 | 2017-09-18 | Apple Inc | Data driven natural language event detection and classification |
DK201670540A1 (en) | 2016-06-11 | 2018-01-08 | Apple Inc | Application integration with a digital assistant |
GB2552723A (en) | 2016-08-03 | 2018-02-07 | Cirrus Logic Int Semiconductor Ltd | Speaker recognition |
GB2552722A (en) | 2016-08-03 | 2018-02-07 | Cirrus Logic Int Semiconductor Ltd | Speaker recognition |
US10043516B2 (en) * | 2016-09-23 | 2018-08-07 | Apple Inc. | Intelligent automated assistant |
US9959861B2 (en) * | 2016-09-30 | 2018-05-01 | Robert Bosch Gmbh | System and method for speech recognition |
US10593346B2 (en) | 2016-12-22 | 2020-03-17 | Apple Inc. | Rank-reduced token representation for automatic speech recognition |
KR20180076830A (en) * | 2016-12-28 | 2018-07-06 | 엘지전자 주식회사 | Audio device and method for controlling the same |
US10229685B2 (en) * | 2017-01-18 | 2019-03-12 | International Business Machines Corporation | Symbol sequence estimation in speech |
CN110235087B (en) * | 2017-01-20 | 2021-06-08 | 华为技术有限公司 | Method and terminal for realizing voice control |
DK180048B1 (en) | 2017-05-11 | 2020-02-04 | Apple Inc. | MAINTAINING THE DATA PROTECTION OF PERSONAL INFORMATION |
DK201770439A1 (en) | 2017-05-11 | 2018-12-13 | Apple Inc. | Offline personal assistant |
US10726832B2 (en) | 2017-05-11 | 2020-07-28 | Apple Inc. | Maintaining privacy of personal information |
DK179496B1 (en) | 2017-05-12 | 2019-01-15 | Apple Inc. | USER-SPECIFIC Acoustic Models |
DK179745B1 (en) | 2017-05-12 | 2019-05-01 | Apple Inc. | SYNCHRONIZATION AND TASK DELEGATION OF A DIGITAL ASSISTANT |
DK201770429A1 (en) | 2017-05-12 | 2018-12-14 | Apple Inc. | Low-latency intelligent automated assistant |
DK201770432A1 (en) | 2017-05-15 | 2018-12-21 | Apple Inc. | Hierarchical belief states for digital assistants |
DK201770411A1 (en) | 2017-05-15 | 2018-12-20 | Apple Inc. | Multi-modal interfaces |
DK201770431A1 (en) | 2017-05-15 | 2018-12-20 | Apple Inc. | Optimizing dialogue policy decisions for digital assistants using implicit feedback |
US20180336892A1 (en) | 2017-05-16 | 2018-11-22 | Apple Inc. | Detecting a trigger of a digital assistant |
DK179560B1 (en) | 2017-05-16 | 2019-02-18 | Apple Inc. | Far-field extension for digital assistant services |
US20180336275A1 (en) | 2017-05-16 | 2018-11-22 | Apple Inc. | Intelligent automated assistant for media exploration |
US11056104B2 (en) * | 2017-05-26 | 2021-07-06 | International Business Machines Corporation | Closed captioning through language detection |
US10650802B2 (en) * | 2017-07-05 | 2020-05-12 | Panasonic Intellectual Property Management Co., Ltd. | Voice recognition method, recording medium, voice recognition device, and robot |
US10657327B2 (en) * | 2017-08-01 | 2020-05-19 | International Business Machines Corporation | Dynamic homophone/synonym identification and replacement for natural language processing |
US10957313B1 (en) * | 2017-09-22 | 2021-03-23 | Amazon Technologies, Inc. | System command processing |
US10600419B1 (en) | 2017-09-22 | 2020-03-24 | Amazon Technologies, Inc. | System command processing |
US10515640B2 (en) * | 2017-11-08 | 2019-12-24 | Intel Corporation | Generating dialogue based on verification scores |
US10818288B2 (en) | 2018-03-26 | 2020-10-27 | Apple Inc. | Natural assistant interaction |
US10943606B2 (en) | 2018-04-12 | 2021-03-09 | Qualcomm Incorporated | Context-based detection of end-point of utterance |
US11145294B2 (en) | 2018-05-07 | 2021-10-12 | Apple Inc. | Intelligent automated assistant for delivering content from user experiences |
US10928918B2 (en) | 2018-05-07 | 2021-02-23 | Apple Inc. | Raise to speak |
US10635462B2 (en) * | 2018-05-23 | 2020-04-28 | Microsoft Technology Licensing, Llc | Skill discovery for computerized personal assistant |
US10892996B2 (en) | 2018-06-01 | 2021-01-12 | Apple Inc. | Variable latency device coordination |
DK180639B1 (en) | 2018-06-01 | 2021-11-04 | Apple Inc | DISABILITY OF ATTENTION-ATTENTIVE VIRTUAL ASSISTANT |
DK179822B1 (en) | 2018-06-01 | 2019-07-12 | Apple Inc. | Voice interaction at a primary device to access call functionality of a companion device |
US10832675B2 (en) * | 2018-08-24 | 2020-11-10 | Denso International America, Inc. | Speech recognition system with interactive spelling function |
US11462215B2 (en) | 2018-09-28 | 2022-10-04 | Apple Inc. | Multi-modal inputs for voice commands |
US11348573B2 (en) | 2019-03-18 | 2022-05-31 | Apple Inc. | Multimodality in digital assistant systems |
US10991365B2 (en) * | 2019-04-08 | 2021-04-27 | Microsoft Technology Licensing, Llc | Automated speech recognition confidence classifier |
US11307752B2 (en) | 2019-05-06 | 2022-04-19 | Apple Inc. | User configurable task triggers |
DK201970509A1 (en) | 2019-05-06 | 2021-01-15 | Apple Inc | Spoken notifications |
CN110085261B (en) * | 2019-05-16 | 2021-08-24 | 上海流利说信息技术有限公司 | Pronunciation correction method, device, equipment and computer readable storage medium |
CN110111775B (en) * | 2019-05-17 | 2021-06-22 | 腾讯科技(深圳)有限公司 | Streaming voice recognition method, device, equipment and storage medium |
US11140099B2 (en) | 2019-05-21 | 2021-10-05 | Apple Inc. | Providing message response suggestions |
DK201970510A1 (en) | 2019-05-31 | 2021-02-11 | Apple Inc | Voice identification in digital assistant systems |
DK180129B1 (en) | 2019-05-31 | 2020-06-02 | Apple Inc. | User activity shortcut suggestions |
US11227599B2 (en) | 2019-06-01 | 2022-01-18 | Apple Inc. | Methods and user interfaces for voice-based control of electronic devices |
CN110718223B (en) * | 2019-10-28 | 2021-02-12 | 百度在线网络技术(北京)有限公司 | Method, apparatus, device and medium for voice interaction control |
US11232786B2 (en) * | 2019-11-27 | 2022-01-25 | Disney Enterprises, Inc. | System and method to improve performance of a speech recognition system by measuring amount of confusion between words |
CN110930988B (en) * | 2019-12-13 | 2020-10-20 | 广州三人行壹佰教育科技有限公司 | Method and system for determining phoneme score |
CN111341305B (en) * | 2020-03-05 | 2023-09-26 | 苏宁云计算有限公司 | Audio data labeling method, device and system |
US11061543B1 (en) | 2020-05-11 | 2021-07-13 | Apple Inc. | Providing relevant data items based on context |
US11183193B1 (en) | 2020-05-11 | 2021-11-23 | Apple Inc. | Digital assistant hardware abstraction |
US11755276B2 (en) | 2020-05-12 | 2023-09-12 | Apple Inc. | Reducing description length based on confidence |
US11908450B2 (en) * | 2020-05-26 | 2024-02-20 | Servicenow, Inc. | Dynamic translation for a conversation |
US20210375270A1 (en) * | 2020-06-02 | 2021-12-02 | Knowles Electronics, Llc | Methods and systems for confusion reduction for compressed acoustic models |
US11490204B2 (en) | 2020-07-20 | 2022-11-01 | Apple Inc. | Multi-device audio adjustment coordination |
US11438683B2 (en) | 2020-07-21 | 2022-09-06 | Apple Inc. | User identification using headphones |
US12008048B2 (en) * | 2021-01-04 | 2024-06-11 | Oracle International Corporation | Drill back to original audio clip in virtual assistant initiated lists and reminders |
KR20230013826A (en) * | 2021-07-20 | 2023-01-27 | 울산과학기술원 | Method and device for providing voice recognition service using artificial intelligence model |
CN113793604B (en) * | 2021-09-14 | 2024-01-05 | 思必驰科技股份有限公司 | Speech recognition system optimization method and device |
Citations (23)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5638425A (en) * | 1992-12-17 | 1997-06-10 | Bell Atlantic Network Services, Inc. | Automated directory assistance system using word recognition and phoneme processing method |
US5895447A (en) * | 1996-02-02 | 1999-04-20 | International Business Machines Corporation | Speech recognition using thresholded speaker class model selection or model adaptation |
US5963903A (en) * | 1996-06-28 | 1999-10-05 | Microsoft Corporation | Method and system for dynamically adjusted training for speech recognition |
US6088669A (en) * | 1997-01-28 | 2000-07-11 | International Business Machines, Corporation | Speech recognition with attempted speaker recognition for speaker model prefetching or alternative speech modeling |
US6107935A (en) * | 1998-02-11 | 2000-08-22 | International Business Machines Corporation | Systems and methods for access filtering employing relaxed recognition constraints |
US6192337B1 (en) * | 1998-08-14 | 2001-02-20 | International Business Machines Corporation | Apparatus and methods for rejecting confusible words during training associated with a speech recognition system |
US6405170B1 (en) | 1998-09-22 | 2002-06-11 | Speechworks International, Inc. | Method and system of reviewing the behavior of an interactive speech recognition application |
US20020120447A1 (en) * | 2000-11-07 | 2002-08-29 | Charlesworth Jason Peter Andrew | Speech processing system |
US20030004717A1 (en) | 2001-03-22 | 2003-01-02 | Nikko Strom | Histogram grammar weighting and error corrective training of grammar weights |
US6601029B1 (en) * | 1999-12-11 | 2003-07-29 | International Business Machines Corporation | Voice processing apparatus |
US20030229492A1 (en) * | 2002-06-05 | 2003-12-11 | Nolan Marc Edward | Biometric identification system |
US20050010412A1 (en) * | 2003-07-07 | 2005-01-13 | Hagai Aronowitz | Phoneme lattice construction and its application to speech recognition and keyword spotting |
US20050238207A1 (en) * | 2004-04-23 | 2005-10-27 | Clifford Tavares | Biometric verification system and method utilizing a data classifier and fusion model |
US20060041756A1 (en) * | 2004-08-19 | 2006-02-23 | International Business Machine Corporation | Systems and methods of securing resources through passwords |
US20060143015A1 (en) | 2004-09-16 | 2006-06-29 | Sbc Technology Resources, Inc. | System and method for facilitating call routing using speech recognition |
US7103542B2 (en) * | 2001-12-14 | 2006-09-05 | Ben Franklin Patent Holding Llc | Automatically improving a voice recognition system |
US7117159B1 (en) * | 2001-09-26 | 2006-10-03 | Sprint Spectrum L.P. | Method and system for dynamic control over modes of operation of voice-processing in a voice command platform |
US7146383B2 (en) * | 2002-10-31 | 2006-12-05 | Sbc Properties, L.P. | Method and system for an automated disambiguation |
US7228278B2 (en) * | 2004-07-06 | 2007-06-05 | Voxify, Inc. | Multi-slot dialog systems and methods |
US20080101658A1 (en) * | 2005-12-22 | 2008-05-01 | James Ahern | Biometric authentication system |
US7412393B1 (en) * | 2004-03-01 | 2008-08-12 | At&T Corp. | Method for developing a dialog manager using modular spoken-dialog components |
US7526431B2 (en) * | 2001-09-05 | 2009-04-28 | Voice Signal Technologies, Inc. | Speech recognition using ambiguous or phone key spelling and/or filtering |
US7864987B2 (en) * | 2006-04-18 | 2011-01-04 | Infosys Technologies Ltd. | Methods and systems for secured access to devices and systems |
Family Cites Families (25)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5566272A (en) * | 1993-10-27 | 1996-10-15 | Lucent Technologies Inc. | Automatic speech recognition (ASR) processing using confidence measures |
US6064959A (en) * | 1997-03-28 | 2000-05-16 | Dragon Systems, Inc. | Error correction in speech recognition |
US6601027B1 (en) * | 1995-11-13 | 2003-07-29 | Scansoft, Inc. | Position manipulation in speech recognition |
US5911485A (en) * | 1995-12-11 | 1999-06-15 | Unwired Planet, Inc. | Predictive data entry method for a keypad |
WO1997042626A1 (en) * | 1996-05-03 | 1997-11-13 | British Telecommunications Public Limited Company | Automatic speech recognition |
DE69833987T2 (en) * | 1998-12-17 | 2006-11-16 | Sony Corp. | Semi-monitored speaker adaptation |
US6513004B1 (en) * | 1999-11-24 | 2003-01-28 | Matsushita Electric Industrial Co., Ltd. | Optimized local feature extraction for automatic speech recognition |
DE10046359A1 (en) * | 2000-09-20 | 2002-03-28 | Philips Corp Intellectual Pty | dialog system |
CN1123863C (en) * | 2000-11-10 | 2003-10-08 | 清华大学 | Information check method based on speed recognition |
EP1209106A1 (en) * | 2000-11-23 | 2002-05-29 | Mars B.V. | Conveyor and method for conveying products |
US20020087311A1 (en) * | 2000-12-29 | 2002-07-04 | Leung Lee Victor Wai | Computer-implemented dynamic language model generation method and system |
US7640190B1 (en) * | 2003-06-24 | 2009-12-29 | Evercom Systems, Inc. | Systems and methods for transaction and information management |
US7502737B2 (en) * | 2002-06-24 | 2009-03-10 | Intel Corporation | Multi-pass recognition of spoken dialogue |
US7386454B2 (en) * | 2002-07-31 | 2008-06-10 | International Business Machines Corporation | Natural error handling in speech recognition |
CN1301026C (en) * | 2002-12-19 | 2007-02-14 | 北京天朗语音科技有限公司 | Mobile terminal voice telephone directory system |
KR100577387B1 (en) * | 2003-08-06 | 2006-05-10 | 삼성전자주식회사 | Method and apparatus for handling speech recognition errors in spoken dialogue systems |
US20050080628A1 (en) * | 2003-10-10 | 2005-04-14 | Metaphor Solutions, Inc. | System, method, and programming language for developing and running dialogs between a user and a virtual agent |
GB2409087A (en) * | 2003-12-12 | 2005-06-15 | Ibm | Computer generated prompting |
US8160883B2 (en) * | 2004-01-10 | 2012-04-17 | Microsoft Corporation | Focus tracking in dialogs |
US8589156B2 (en) * | 2004-07-12 | 2013-11-19 | Hewlett-Packard Development Company, L.P. | Allocation of speech recognition tasks and combination of results thereof |
US8005668B2 (en) * | 2004-09-22 | 2011-08-23 | General Motors Llc | Adaptive confidence thresholds in telematics system speech recognition |
US7716056B2 (en) * | 2004-09-27 | 2010-05-11 | Robert Bosch Corporation | Method and system for interactive conversational dialogue for cognitively overloaded device users |
GB0426347D0 (en) | 2004-12-01 | 2005-01-05 | Ibm | Methods, apparatus and computer programs for automatic speech recognition |
US7809569B2 (en) * | 2004-12-22 | 2010-10-05 | Enterprise Integration Group, Inc. | Turn-taking confidence |
US20080208586A1 (en) | 2007-02-27 | 2008-08-28 | Soonthorn Ativanichayaphong | Enabling Natural Language Understanding In An X+V Page Of A Multimodal Application |
-
2004
- 2004-12-01 GB GBGB0426347.1A patent/GB0426347D0/en not_active Ceased
-
2005
- 2005-10-20 US US11/255,201 patent/US8694316B2/en active Active
- 2005-11-11 CN CN2005101149067A patent/CN1783213B/en not_active Expired - Fee Related
-
2014
- 2014-02-26 US US14/191,176 patent/US9502024B2/en active Active
Patent Citations (23)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5638425A (en) * | 1992-12-17 | 1997-06-10 | Bell Atlantic Network Services, Inc. | Automated directory assistance system using word recognition and phoneme processing method |
US5895447A (en) * | 1996-02-02 | 1999-04-20 | International Business Machines Corporation | Speech recognition using thresholded speaker class model selection or model adaptation |
US5963903A (en) * | 1996-06-28 | 1999-10-05 | Microsoft Corporation | Method and system for dynamically adjusted training for speech recognition |
US6088669A (en) * | 1997-01-28 | 2000-07-11 | International Business Machines, Corporation | Speech recognition with attempted speaker recognition for speaker model prefetching or alternative speech modeling |
US6107935A (en) * | 1998-02-11 | 2000-08-22 | International Business Machines Corporation | Systems and methods for access filtering employing relaxed recognition constraints |
US6192337B1 (en) * | 1998-08-14 | 2001-02-20 | International Business Machines Corporation | Apparatus and methods for rejecting confusible words during training associated with a speech recognition system |
US6405170B1 (en) | 1998-09-22 | 2002-06-11 | Speechworks International, Inc. | Method and system of reviewing the behavior of an interactive speech recognition application |
US6601029B1 (en) * | 1999-12-11 | 2003-07-29 | International Business Machines Corporation | Voice processing apparatus |
US20020120447A1 (en) * | 2000-11-07 | 2002-08-29 | Charlesworth Jason Peter Andrew | Speech processing system |
US20030004717A1 (en) | 2001-03-22 | 2003-01-02 | Nikko Strom | Histogram grammar weighting and error corrective training of grammar weights |
US7526431B2 (en) * | 2001-09-05 | 2009-04-28 | Voice Signal Technologies, Inc. | Speech recognition using ambiguous or phone key spelling and/or filtering |
US7117159B1 (en) * | 2001-09-26 | 2006-10-03 | Sprint Spectrum L.P. | Method and system for dynamic control over modes of operation of voice-processing in a voice command platform |
US7103542B2 (en) * | 2001-12-14 | 2006-09-05 | Ben Franklin Patent Holding Llc | Automatically improving a voice recognition system |
US20030229492A1 (en) * | 2002-06-05 | 2003-12-11 | Nolan Marc Edward | Biometric identification system |
US7146383B2 (en) * | 2002-10-31 | 2006-12-05 | Sbc Properties, L.P. | Method and system for an automated disambiguation |
US20050010412A1 (en) * | 2003-07-07 | 2005-01-13 | Hagai Aronowitz | Phoneme lattice construction and its application to speech recognition and keyword spotting |
US7412393B1 (en) * | 2004-03-01 | 2008-08-12 | At&T Corp. | Method for developing a dialog manager using modular spoken-dialog components |
US20050238207A1 (en) * | 2004-04-23 | 2005-10-27 | Clifford Tavares | Biometric verification system and method utilizing a data classifier and fusion model |
US7228278B2 (en) * | 2004-07-06 | 2007-06-05 | Voxify, Inc. | Multi-slot dialog systems and methods |
US20060041756A1 (en) * | 2004-08-19 | 2006-02-23 | International Business Machine Corporation | Systems and methods of securing resources through passwords |
US20060143015A1 (en) | 2004-09-16 | 2006-06-29 | Sbc Technology Resources, Inc. | System and method for facilitating call routing using speech recognition |
US20080101658A1 (en) * | 2005-12-22 | 2008-05-01 | James Ahern | Biometric authentication system |
US7864987B2 (en) * | 2006-04-18 | 2011-01-04 | Infosys Technologies Ltd. | Methods and systems for secured access to devices and systems |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US9502024B2 (en) | 2004-12-01 | 2016-11-22 | Nuance Communications, Inc. | Methods, apparatus and computer programs for automatic speech recognition |
US10102771B2 (en) | 2013-04-26 | 2018-10-16 | Wistron Corporation | Method and device for learning language and computer readable recording medium |
US9953646B2 (en) | 2014-09-02 | 2018-04-24 | Belleau Technologies | Method and system for dynamic speech recognition and tracking of prewritten script |
Also Published As
Publication number | Publication date |
---|---|
US20060116877A1 (en) | 2006-06-01 |
CN1783213B (en) | 2010-06-09 |
US9502024B2 (en) | 2016-11-22 |
US20140249816A1 (en) | 2014-09-04 |
CN1783213A (en) | 2006-06-07 |
GB0426347D0 (en) | 2005-01-05 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US9502024B2 (en) | Methods, apparatus and computer programs for automatic speech recognition | |
US6925154B2 (en) | Methods and apparatus for conversational name dialing systems | |
KR100383353B1 (en) | Speech recognition apparatus and method of generating vocabulary for the same | |
US7392188B2 (en) | System and method enabling acoustic barge-in | |
US6601029B1 (en) | Voice processing apparatus | |
US7437291B1 (en) | Using partial information to improve dialog in automatic speech recognition systems | |
US8914294B2 (en) | System and method of providing an automated data-collection in spoken dialog systems | |
US5995928A (en) | Method and apparatus for continuous spelling speech recognition with early identification | |
US6487530B1 (en) | Method for recognizing non-standard and standard speech by speaker independent and speaker dependent word models | |
US20160372116A1 (en) | Voice authentication and speech recognition system and method | |
CA2486125C (en) | A system and method of using meta-data in speech-processing | |
US6438520B1 (en) | Apparatus, method and system for cross-speaker speech recognition for telecommunication applications | |
US7533018B2 (en) | Tailored speaker-independent voice recognition system | |
US8428944B2 (en) | System and method for performing compensated speech recognition | |
JPH08234788A (en) | Method and equipment for bias equalization of speech recognition | |
US7447636B1 (en) | System and methods for using transcripts to train an automated directory assistance service | |
CN110570853A (en) | Intention recognition method and device based on voice data | |
USH2187H1 (en) | System and method for gender identification in a speech application environment | |
US7401023B1 (en) | Systems and methods for providing automated directory assistance using transcripts | |
US20040098259A1 (en) | Method for recognition verbal utterances by a non-mother tongue speaker in a speech processing system | |
JP2005534983A (en) | Automatic speech recognition method | |
US20010056345A1 (en) | Method and system for speech recognition of the alphabet | |
KR100622019B1 (en) | Voice interface system and method | |
US20080243498A1 (en) | Method and system for providing interactive speech recognition using speaker data | |
Juang et al. | Deployable automatic speech recognition systems: Advances and challenges |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
AS | Assignment |
Owner name: INTERNATIONAL BUSINESS MACHINES CORPORATION, NEW Y Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNORS:PICKERING, JOHN BRIAN;POULTNEY, TIMOTHY DAVID;STANIFORD, BENJAMIN TERRICK;AND OTHERS;REEL/FRAME:017026/0165;SIGNING DATES FROM 20051007 TO 20051012 Owner name: INTERNATIONAL BUSINESS MACHINES CORPORATION, NEW Y Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNORS:PICKERING, JOHN BRIAN;POULTNEY, TIMOTHY DAVID;STANIFORD, BENJAMIN TERRICK;AND OTHERS;SIGNING DATES FROM 20051007 TO 20051012;REEL/FRAME:017026/0165 |
|
AS | Assignment |
Owner name: NUANCE COMMUNICATIONS, INC., MASSACHUSETTS Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:INTERNATIONAL BUSINESS MACHINES CORPORATION;REEL/FRAME:022689/0317 Effective date: 20090331 Owner name: NUANCE COMMUNICATIONS, INC.,MASSACHUSETTS Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:INTERNATIONAL BUSINESS MACHINES CORPORATION;REEL/FRAME:022689/0317 Effective date: 20090331 |
|
STCF | Information on status: patent grant |
Free format text: PATENTED CASE |
|
MAFP | Maintenance fee payment |
Free format text: PAYMENT OF MAINTENANCE FEE, 4TH YEAR, LARGE ENTITY (ORIGINAL EVENT CODE: M1551) Year of fee payment: 4 |
|
MAFP | Maintenance fee payment |
Free format text: PAYMENT OF MAINTENANCE FEE, 8TH YEAR, LARGE ENTITY (ORIGINAL EVENT CODE: M1552); ENTITY STATUS OF PATENT OWNER: LARGE ENTITY Year of fee payment: 8 |
|
AS | Assignment |
Owner name: MICROSOFT TECHNOLOGY LICENSING, LLC, WASHINGTON Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:NUANCE COMMUNICATIONS, INC.;REEL/FRAME:065530/0871 Effective date: 20230920 |